Technology executive · Applied AI in healthcare

AI that works
in production,
in healthcare.

I’m Jay Leon. For twenty years my job has been the last mile: taking a platform and making it work inside someone else’s reality. Today I’m CTO at LifeChef Health, where I built a HIPAA-grade clinical platform, own the partner integrations, and lead AI adoption across operations.

Based in Northwest Arkansas. Leading a distributed engineering team.

Where I’ve shippedLifeChef HealthFlashpointKBS+The New York TimesBanana Republic

Selected work

The last mile is the job.

A platform that works in a demo is the easy part. These are three times I made one work inside a real organization, with real constraints and people depending on the result.

Healthcare · CTO, LifeChef Health

3 weeks to deliver the core of a HIPAA-compliant clinical platform, with a team of four

One patient record, from referral to delivery

Situation
Medically tailored meals only work when referrals, the EHR, ordering, the kitchen and delivery agree on the same patient. They lived in separate systems.
What I did
I designed and built the Member Intelligence Platform, connecting the Canvas EHR, CRM, ordering, kitchen production and fulfillment to HIPAA rules. I own the integrations with our partners, including a national community referral network and regional care organizations.
What changed
The core shipped in three weeks with four engineers. It runs on a HIPAA-ready cloud backbone at 99.9% uptime, with cloud costs down 40%.

Healthcare operations · CTO, LifeChef Health

3 days → 5 hrs to resolve a payment dispute or refund, once the workflow was automated

AI adoption across operations

Situation
Finance, kitchen, warehouse and clinical teams spent their days on repetitive work that software could carry, but none of it could fail quietly.
What I did
I run LifeChef’s AI adoption program. Each workflow gets a written spec and acceptance criteria before anything is automated, and people keep sign-off where judgment matters.
What changed
Automation now runs in finance, the kitchen, the warehouse and clinical operations. In finance alone, dispute and refund resolution went from three days to five hours, saving about $500K a year. The teams spend their time on the exceptions instead of the routine.

Threat intelligence · Architect to VP of Technology, 2013–2015

$2M+ in revenue growth enabled by the platform I led

Intelligence for customers who can’t accept “almost”

Situation
Flashpoint delivered intelligence to the FBI, CIA, DOD, Visa, Mastercard and JPMorgan Chase. None of them tolerate a deployment that almost works.
What I did
I was founding architect of the core stack, then moved from Lead Platform Architect to Director and VP of Technology, responsible for the platform and the delivery around it.
What changed
The deep web intelligence platform became the product the company sold. It enabled more than $2M in revenue growth and strategic investment from TechOperators, Greycroft Partners and Cisco Investments.

He’s built systems large and small that continue to scale … in some cases going in and re-architecting failing systems that used fewer resources to achieve greater results.

Mike DorySr. Staff UX Engineering Lead, Google DeepMind · via LinkedIn

Having Jay on your team is sure to give you confidence that even the craziest brilliant ideas can become a reality.

Marc HartzmanExecutive Creative Director, KBS+ · via LinkedIn

Track record

  1. LifeChef HealthNow

    Chief Technology Officer2020 – now

    Built the HIPAA-grade clinical platform, own the partner integrations, lead AI adoption across operations.

  2. Front Row LabsNow

    CTO / Principal2019 – now

    Technical due diligence and post-investment execution: the bridge between investment capital and engineering reality.

  3. FreeD Group

    CTO, Technology Architect2016 – 2018

    Led an AI-powered travel assistant backed by 500 Startups, years before the current wave.

  4. Flashpoint

    VP of Technology2013 – 2015

    Founding architect of the core stack, then Director and VP. Intelligence products for government agencies and the largest banks.

  5. Spies & Assassins / KBS+

    Lead Platform Engineer2010 – 2013

    Third engineering hire at the agency’s innovation lab. Scaled the team from 3 to more than 40.

  6. WEBApps.NYC

    Founder, fractional CTO2010 – 2021

    Eleven years working inside organizations like The New York Times and Banana Republic.

The full history is on LinkedIn.

How deployment gets repeatable

No spec, no work.

Most deployments that fail, fail before the code: in a vague scope, an unstated assumption about the customer’s systems, or a finish line nobody wrote down.

  1. Write it down

    Each deployment starts with a written description of the problem and how we’ll know it’s solved.

  2. Build against it

    The build is checked against that document, not against a demo. Same scoping, same acceptance criteria, every time.

  3. Ship when it matches

    Nothing ships until it matches, and what the field learns flows back into the product.

I call the discipline Enso. It matters most with AI, where a system can look finished long before it is.

Who I am

Jay Leon

I’m an engineer who became an executive and never stopped building.

For eleven years I ran WEBApps.NYC as a fractional CTO, working inside organizations like The New York Times and Banana Republic. At KBS+ I was the third engineering hire in the innovation lab and scaled the team from 3 to more than 40.

At Flashpoint I was founding architect of the core stack and went on to be VP of Technology, delivering intelligence products to government agencies and the largest banks.

Since 2020 I’ve been CTO at LifeChef Health, taking it from a meal-delivery startup to a clinically integrated platform and working directly with registered dietitians and medical advisors. I lead a distributed team of four engineers there, and I’ve done the job from both sides. I’ve been the vendor trying to land inside a customer’s systems, and the customer deciding whether a vendor’s platform is real. That is the experience I bring to a deployment team.

Common questions

What do you mean by “the last mile”?

The distance between a platform that works and a platform that works inside a specific organization: its systems, its data, its rules and its people. Most of the risk in a deployment lives there, and most of my career has been spent there.

Why healthcare?

Because the constraints are real. EHR integration, HIPAA and clinical workflow punish shortcuts, and the cost of a failure lands on patients and clinicians. I’ve built and run clinical systems under those rules, which is a different thing from reading about them.

What is Enso?

Enso is the spec-first discipline I use for AI-assisted work: no spec, no work. Every piece of work starts with a written problem statement and acceptance criteria, and is checked against them before it ships. The full methodology is on the Enso page.

Do you still write code?

Yes, deliberately. When a decision depends on whether something can actually be built, such as an AI workflow or a hard integration, I often build the first version myself. It’s the quickest way to replace opinion with evidence. Some of it is public: WebFetch.MCP, my open-source MCP server for giving AI clients web search and content extraction, is on GitHub.

Where are you based?

Northwest Arkansas. I work remotely and have led distributed engineering teams across time zones for years.

Why I do this work

Nothing we build stands alone.

Everything I build depends on the people it reaches: a patient waiting on a result, a clinician at the end of a long shift, an engineer who deserves to go home on time.

So I build with care for all of them: clear specifications instead of guesswork, less waste, fewer surprises, and technology that eases suffering rather than adding to it.

Let’s talk.

If you’re putting AI into production in healthcare, or anywhere a deployment that almost works isn’t good enough, I’d like to hear about it. A 15-minute call is a good start.

Book a call

Prefer to write first? Message me on LinkedIn.